|
rpm-build |
d9acb6 |
import unittest, sys
|
|
rpm-build |
d9acb6 |
from lxml.tests.common_imports import make_doctest
|
|
rpm-build |
d9acb6 |
from lxml.etree import LIBXML_VERSION
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
import lxml.html
|
|
rpm-build |
d9acb6 |
from lxml.html.clean import Cleaner, clean_html
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
class CleanerTest(unittest.TestCase):
|
|
rpm-build |
d9acb6 |
def test_allow_tags(self):
|
|
rpm-build |
d9acb6 |
html = """
|
|
rpm-build |
d9acb6 |
<html>
|
|
rpm-build |
d9acb6 |
<head>
|
|
rpm-build |
d9acb6 |
</head>
|
|
rpm-build |
d9acb6 |
<body>
|
|
rpm-build |
d9acb6 |
some text
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
helloworld
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
helloworld
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
![]()
|
|
rpm-build |
d9acb6 |
</body>
|
|
rpm-build |
d9acb6 |
</html>
|
|
rpm-build |
d9acb6 |
"""
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
html_root = lxml.html.document_fromstring(html)
|
|
rpm-build |
d9acb6 |
cleaner = Cleaner(
|
|
rpm-build |
d9acb6 |
remove_unknown_tags = False,
|
|
rpm-build |
d9acb6 |
allow_tags = ['table', 'tr', 'td'])
|
|
rpm-build |
d9acb6 |
result = cleaner.clean_html(html_root)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
self.assertEqual(12-5+1, len(list(result.iter())))
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def test_safe_attrs_included(self):
|
|
rpm-build |
d9acb6 |
html = """Cyan """
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
safe_attrs=set(lxml.html.defs.safe_attrs)
|
|
rpm-build |
d9acb6 |
safe_attrs.add('style')
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
cleaner = Cleaner(
|
|
rpm-build |
d9acb6 |
safe_attrs_only=True,
|
|
rpm-build |
d9acb6 |
safe_attrs=safe_attrs)
|
|
rpm-build |
d9acb6 |
result = cleaner.clean_html(html)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
self.assertEqual(html, result)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def test_safe_attrs_excluded(self):
|
|
rpm-build |
d9acb6 |
html = """Cyan """
|
|
rpm-build |
d9acb6 |
expected = """Cyan """
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
safe_attrs=set()
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
cleaner = Cleaner(
|
|
rpm-build |
d9acb6 |
safe_attrs_only=True,
|
|
rpm-build |
d9acb6 |
safe_attrs=safe_attrs)
|
|
rpm-build |
d9acb6 |
result = cleaner.clean_html(html)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
self.assertEqual(expected, result)
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def test_clean_invalid_root_tag(self):
|
|
rpm-build |
d9acb6 |
# only testing that cleaning with invalid root tags works at all
|
|
rpm-build |
d9acb6 |
s = lxml.html.fromstring('parent <invalid tag>child</another>')
|
|
rpm-build |
d9acb6 |
self.assertEqual('parent child', clean_html(s).text_content())
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
s = lxml.html.fromstring('<invalid tag>child</another>')
|
|
rpm-build |
d9acb6 |
self.assertEqual('child', clean_html(s).text_content())
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
|
|
rpm-build |
d9acb6 |
def test_suite():
|
|
rpm-build |
d9acb6 |
suite = unittest.TestSuite()
|
|
rpm-build |
d9acb6 |
if sys.version_info >= (2,4):
|
|
rpm-build |
d9acb6 |
suite.addTests([make_doctest('test_clean.txt')])
|
|
rpm-build |
d9acb6 |
if LIBXML_VERSION >= (2,6,31):
|
|
rpm-build |
d9acb6 |
suite.addTests([make_doctest('test_clean_embed.txt')])
|
|
rpm-build |
d9acb6 |
suite.addTests(unittest.makeSuite(CleanerTest))
|
|
rpm-build |
d9acb6 |
return suite
|